INDEX

Explanations

references to winning or victory

oai_token-act-pair · gpt-3.5-turbo

terminology related to victories and wins

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 intestine

-0.71

 encyclopedia

-0.70

 bowel

-0.70

erity

-0.68

Uz

-0.68

 incorpor

-0.65

 notor

-0.65

 trave

-0.65

 pronouns

-0.62

umn

-0.62

POSITIVE LOGITS

nings

1.22

now

0.95

iem

0.95

ners

0.85

athon

0.82

throp

0.80

iors

0.79

ces

0.78

hardt

0.78

win

0.77

Activations Density 0.023%

references to winning or victory

terminology related to victories and wins

No Comments

No Known Activations